Evidence and limits
Finding: ChatGPT access reduced completion time by about 40% and raised rated output quality by about 18%, while compressing performance differences between participants.
Impact: Strengthen matched time-and-quality measurement. Watch skill compression and task-boundary effects.
Boundary: The tasks were short, self-contained, and text based. Next: do gains survive multi-stage work with persistent context, verification, and accountability?
Flags: short writing tasks; no maintenance outcome
Record State: Catalog review recorded; not a performance endorsement
What was reviewed: Structure and relevance screened; not a full evaluation